CERTAIN TAXONOMIC CHARACTERS OF GRYLLUS. 19 



the base of the spit (average, 8.48) and least in the one from the apex 

 (average, 7.76). Since the means on the mainland and at the apex of 

 the spit markedly differ, we would expect that the variability of the 

 crickets occupying the middle ground would be greater than that of 

 either of the extremes. This is on the supposition that the crickets at 

 the base of the spit are a mixture of migrants from the two extreme 

 populations and also of the population proper to that area. However, 

 crickets do not move about much, and it may be that the greater vari- 

 ability at the base is connected with the greater variation in the consti- 

 tution of the soil and other factors of the environment. A third idea 

 is that optimum conditions favor great variability. Judging from the 

 relative abundance of individuals, the most favorable of the three areas 

 for Gryllus is the base and the least favorable is the apex of the spit. 

 Lastly, selection may be instrumental in reducing the variation at the 

 latter place. Tower, Crampton, Bumpus, and Weldon have shown that 

 selection acts this way. Doubtless many crickets are killed by other 

 animals, but it is difficult to see how there would be much selection 

 in this killing. However, the winter is passed in the egg state, the 

 eggs being laid singly in the ground. Wherever the soil is loose- 

 as on the sand-spit, especially at the apex those eggs which are not 

 deeply placed will almost certainly perish. In this way selection acts 

 against the offspring of females having short ovipositors where the 

 soil is loose. It is to be noted that the length of the ovipositor increases 

 as the looseness of the soil increases. A small collection from the 

 middle of the spit had an average ovipositor-length of 19.52 mm. or 

 about midway between the lengths of the ovipositor of the base and 

 of that of the apex populations, while those from the mainland had the 

 shortest ovipositors. 



The few Gryllus collected in the grass-fields about Perkins Cove, 

 Maine, had an average ovipositor-length of 14 mm., while of the 147 

 females from the cove collection (see p. 27), which had passed the 

 winter as eggs in the beach-sand, only 2 had so small an ovipositor and 

 the average was nearly 19 mm. Which of these four suggestions, if 

 any, explains the differences of variability at Cold Spring Harbor can, 

 at present, be nothing more than a matter of discussion. It is difficult 

 to obtain conclusive experimental evidence. Pearson (1902) states 

 that selection decreases correlation. In these collections the average 

 coefficient of correlation between the various organs here considered is 

 0.70 at the apex as compared with 0.74 at the base and 0.73 on the 

 mainland. To take a specific organ, the only one showing progressive 

 differences is the ovipositor. It has an average coefficient of correla- 

 tion of 0.64 at the apex, 0.67 at the base, and 0.72 on the mainland. 

 This is, furthermore, the organ in which we would expect progressive 



